Intelligent Malaysian Sign Language Translation System Using Convolutional-Based Attention Module with Residual Network

Table 2

Comparison table for CBAM-2DResNet “within blocks” and “before classifier.”

Metrics	CBAM-2DResNet “within blocks”	CBAM-2DResNet “before classifier”
Training duration	16040.79 seconds	4860.23 seconds
Lowest validation loss	0.0214	0.0210
Highest validation accuracy	99.29%	99.39%
F1-score achieved	0.99	0.99
Generalisation performance	Good fit	Good fit